blob: 98a817b537e273baa2b94b3910d723483714d3f8
1 | /* |
2 | * AAC encoder wrapper |
3 | * Copyright (c) 2012 Martin Storsjo |
4 | * |
5 | * This file is part of FFmpeg. |
6 | * |
7 | * Permission to use, copy, modify, and/or distribute this software for any |
8 | * purpose with or without fee is hereby granted, provided that the above |
9 | * copyright notice and this permission notice appear in all copies. |
10 | * |
11 | * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES |
12 | * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF |
13 | * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR |
14 | * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES |
15 | * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN |
16 | * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF |
17 | * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. |
18 | */ |
19 | |
20 | #include <fdk-aac/aacenc_lib.h> |
21 | |
22 | #include "libavutil/channel_layout.h" |
23 | #include "libavutil/common.h" |
24 | #include "libavutil/opt.h" |
25 | #include "avcodec.h" |
26 | #include "audio_frame_queue.h" |
27 | #include "internal.h" |
28 | |
29 | typedef struct AACContext { |
30 | const AVClass *class; |
31 | HANDLE_AACENCODER handle; |
32 | int afterburner; |
33 | int eld_sbr; |
34 | int signaling; |
35 | int latm; |
36 | int header_period; |
37 | int vbr; |
38 | |
39 | AudioFrameQueue afq; |
40 | } AACContext; |
41 | |
42 | static const AVOption aac_enc_options[] = { |
43 | { "afterburner", "Afterburner (improved quality)", offsetof(AACContext, afterburner), AV_OPT_TYPE_INT, { .i64 = 1 }, 0, 1, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM }, |
44 | { "eld_sbr", "Enable SBR for ELD (for SBR in other configurations, use the -profile parameter)", offsetof(AACContext, eld_sbr), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM }, |
45 | { "signaling", "SBR/PS signaling style", offsetof(AACContext, signaling), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 2, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM, "signaling" }, |
46 | { "default", "Choose signaling implicitly (explicit hierarchical by default, implicit if global header is disabled)", 0, AV_OPT_TYPE_CONST, { .i64 = -1 }, 0, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM, "signaling" }, |
47 | { "implicit", "Implicit backwards compatible signaling", 0, AV_OPT_TYPE_CONST, { .i64 = 0 }, 0, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM, "signaling" }, |
48 | { "explicit_sbr", "Explicit SBR, implicit PS signaling", 0, AV_OPT_TYPE_CONST, { .i64 = 1 }, 0, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM, "signaling" }, |
49 | { "explicit_hierarchical", "Explicit hierarchical signaling", 0, AV_OPT_TYPE_CONST, { .i64 = 2 }, 0, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM, "signaling" }, |
50 | { "latm", "Output LATM/LOAS encapsulated data", offsetof(AACContext, latm), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM }, |
51 | { "header_period", "StreamMuxConfig and PCE repetition period (in frames)", offsetof(AACContext, header_period), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 0xffff, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM }, |
52 | { "vbr", "VBR mode (1-5)", offsetof(AACContext, vbr), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 5, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM }, |
53 | { NULL } |
54 | }; |
55 | |
56 | static const AVClass aac_enc_class = { |
57 | "libfdk_aac", av_default_item_name, aac_enc_options, LIBAVUTIL_VERSION_INT |
58 | }; |
59 | |
60 | static const char *aac_get_error(AACENC_ERROR err) |
61 | { |
62 | switch (err) { |
63 | case AACENC_OK: |
64 | return "No error"; |
65 | case AACENC_INVALID_HANDLE: |
66 | return "Invalid handle"; |
67 | case AACENC_MEMORY_ERROR: |
68 | return "Memory allocation error"; |
69 | case AACENC_UNSUPPORTED_PARAMETER: |
70 | return "Unsupported parameter"; |
71 | case AACENC_INVALID_CONFIG: |
72 | return "Invalid config"; |
73 | case AACENC_INIT_ERROR: |
74 | return "Initialization error"; |
75 | case AACENC_INIT_AAC_ERROR: |
76 | return "AAC library initialization error"; |
77 | case AACENC_INIT_SBR_ERROR: |
78 | return "SBR library initialization error"; |
79 | case AACENC_INIT_TP_ERROR: |
80 | return "Transport library initialization error"; |
81 | case AACENC_INIT_META_ERROR: |
82 | return "Metadata library initialization error"; |
83 | case AACENC_ENCODE_ERROR: |
84 | return "Encoding error"; |
85 | case AACENC_ENCODE_EOF: |
86 | return "End of file"; |
87 | default: |
88 | return "Unknown error"; |
89 | } |
90 | } |
91 | |
92 | static int aac_encode_close(AVCodecContext *avctx) |
93 | { |
94 | AACContext *s = avctx->priv_data; |
95 | |
96 | if (s->handle) |
97 | aacEncClose(&s->handle); |
98 | av_freep(&avctx->extradata); |
99 | ff_af_queue_close(&s->afq); |
100 | |
101 | return 0; |
102 | } |
103 | |
104 | static av_cold int aac_encode_init(AVCodecContext *avctx) |
105 | { |
106 | AACContext *s = avctx->priv_data; |
107 | int ret = AVERROR(EINVAL); |
108 | AACENC_InfoStruct info = { 0 }; |
109 | CHANNEL_MODE mode; |
110 | AACENC_ERROR err; |
111 | int aot = FF_PROFILE_AAC_LOW + 1; |
112 | int sce = 0, cpe = 0; |
113 | |
114 | if ((err = aacEncOpen(&s->handle, 0, avctx->channels)) != AACENC_OK) { |
115 | av_log(avctx, AV_LOG_ERROR, "Unable to open the encoder: %s\n", |
116 | aac_get_error(err)); |
117 | goto error; |
118 | } |
119 | |
120 | if (avctx->profile != FF_PROFILE_UNKNOWN) |
121 | aot = avctx->profile + 1; |
122 | |
123 | if ((err = aacEncoder_SetParam(s->handle, AACENC_AOT, aot)) != AACENC_OK) { |
124 | av_log(avctx, AV_LOG_ERROR, "Unable to set the AOT %d: %s\n", |
125 | aot, aac_get_error(err)); |
126 | goto error; |
127 | } |
128 | |
129 | if (aot == FF_PROFILE_AAC_ELD + 1 && s->eld_sbr) { |
130 | if ((err = aacEncoder_SetParam(s->handle, AACENC_SBR_MODE, |
131 | 1)) != AACENC_OK) { |
132 | av_log(avctx, AV_LOG_ERROR, "Unable to enable SBR for ELD: %s\n", |
133 | aac_get_error(err)); |
134 | goto error; |
135 | } |
136 | } |
137 | |
138 | if ((err = aacEncoder_SetParam(s->handle, AACENC_SAMPLERATE, |
139 | avctx->sample_rate)) != AACENC_OK) { |
140 | av_log(avctx, AV_LOG_ERROR, "Unable to set the sample rate %d: %s\n", |
141 | avctx->sample_rate, aac_get_error(err)); |
142 | goto error; |
143 | } |
144 | |
145 | switch (avctx->channels) { |
146 | case 1: mode = MODE_1; sce = 1; cpe = 0; break; |
147 | case 2: mode = MODE_2; sce = 0; cpe = 1; break; |
148 | case 3: mode = MODE_1_2; sce = 1; cpe = 1; break; |
149 | case 4: mode = MODE_1_2_1; sce = 2; cpe = 1; break; |
150 | case 5: mode = MODE_1_2_2; sce = 1; cpe = 2; break; |
151 | case 6: mode = MODE_1_2_2_1; sce = 2; cpe = 2; break; |
152 | /* The version macro is introduced the same time as the 7.1 support, so this |
153 | should suffice. */ |
154 | #ifdef AACENCODER_LIB_VL0 |
155 | case 8: |
156 | sce = 2; |
157 | cpe = 3; |
158 | if (avctx->channel_layout == AV_CH_LAYOUT_7POINT1) { |
159 | mode = MODE_7_1_REAR_SURROUND; |
160 | } else { |
161 | // MODE_1_2_2_2_1 and MODE_7_1_FRONT_CENTER use the same channel layout |
162 | mode = MODE_7_1_FRONT_CENTER; |
163 | } |
164 | break; |
165 | #endif |
166 | default: |
167 | av_log(avctx, AV_LOG_ERROR, |
168 | "Unsupported number of channels %d\n", avctx->channels); |
169 | goto error; |
170 | } |
171 | |
172 | if ((err = aacEncoder_SetParam(s->handle, AACENC_CHANNELMODE, |
173 | mode)) != AACENC_OK) { |
174 | av_log(avctx, AV_LOG_ERROR, |
175 | "Unable to set channel mode %d: %s\n", mode, aac_get_error(err)); |
176 | goto error; |
177 | } |
178 | |
179 | if ((err = aacEncoder_SetParam(s->handle, AACENC_CHANNELORDER, |
180 | 1)) != AACENC_OK) { |
181 | av_log(avctx, AV_LOG_ERROR, |
182 | "Unable to set wav channel order %d: %s\n", |
183 | mode, aac_get_error(err)); |
184 | goto error; |
185 | } |
186 | |
187 | if (avctx->flags & AV_CODEC_FLAG_QSCALE || s->vbr) { |
188 | int mode = s->vbr ? s->vbr : avctx->global_quality; |
189 | if (mode < 1 || mode > 5) { |
190 | av_log(avctx, AV_LOG_WARNING, |
191 | "VBR quality %d out of range, should be 1-5\n", mode); |
192 | mode = av_clip(mode, 1, 5); |
193 | } |
194 | av_log(avctx, AV_LOG_WARNING, |
195 | "Note, the VBR setting is unsupported and only works with " |
196 | "some parameter combinations\n"); |
197 | if ((err = aacEncoder_SetParam(s->handle, AACENC_BITRATEMODE, |
198 | mode)) != AACENC_OK) { |
199 | av_log(avctx, AV_LOG_ERROR, "Unable to set the VBR bitrate mode %d: %s\n", |
200 | mode, aac_get_error(err)); |
201 | goto error; |
202 | } |
203 | } else { |
204 | if (avctx->bit_rate <= 0) { |
205 | if (avctx->profile == FF_PROFILE_AAC_HE_V2) { |
206 | sce = 1; |
207 | cpe = 0; |
208 | } |
209 | avctx->bit_rate = (96*sce + 128*cpe) * avctx->sample_rate / 44; |
210 | if (avctx->profile == FF_PROFILE_AAC_HE || |
211 | avctx->profile == FF_PROFILE_AAC_HE_V2 || |
212 | avctx->profile == FF_PROFILE_MPEG2_AAC_HE || |
213 | s->eld_sbr) |
214 | avctx->bit_rate /= 2; |
215 | } |
216 | if ((err = aacEncoder_SetParam(s->handle, AACENC_BITRATE, |
217 | avctx->bit_rate)) != AACENC_OK) { |
218 | av_log(avctx, AV_LOG_ERROR, "Unable to set the bitrate %"PRId64": %s\n", |
219 | (int64_t)avctx->bit_rate, aac_get_error(err)); |
220 | goto error; |
221 | } |
222 | } |
223 | |
224 | /* Choose bitstream format - if global header is requested, use |
225 | * raw access units, otherwise use ADTS. */ |
226 | if ((err = aacEncoder_SetParam(s->handle, AACENC_TRANSMUX, |
227 | avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER ? 0 : s->latm ? 10 : 2)) != AACENC_OK) { |
228 | av_log(avctx, AV_LOG_ERROR, "Unable to set the transmux format: %s\n", |
229 | aac_get_error(err)); |
230 | goto error; |
231 | } |
232 | |
233 | if (s->latm && s->header_period) { |
234 | if ((err = aacEncoder_SetParam(s->handle, AACENC_HEADER_PERIOD, |
235 | s->header_period)) != AACENC_OK) { |
236 | av_log(avctx, AV_LOG_ERROR, "Unable to set header period: %s\n", |
237 | aac_get_error(err)); |
238 | goto error; |
239 | } |
240 | } |
241 | |
242 | /* If no signaling mode is chosen, use explicit hierarchical signaling |
243 | * if using mp4 mode (raw access units, with global header) and |
244 | * implicit signaling if using ADTS. */ |
245 | if (s->signaling < 0) |
246 | s->signaling = avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER ? 2 : 0; |
247 | |
248 | if ((err = aacEncoder_SetParam(s->handle, AACENC_SIGNALING_MODE, |
249 | s->signaling)) != AACENC_OK) { |
250 | av_log(avctx, AV_LOG_ERROR, "Unable to set signaling mode %d: %s\n", |
251 | s->signaling, aac_get_error(err)); |
252 | goto error; |
253 | } |
254 | |
255 | if ((err = aacEncoder_SetParam(s->handle, AACENC_AFTERBURNER, |
256 | s->afterburner)) != AACENC_OK) { |
257 | av_log(avctx, AV_LOG_ERROR, "Unable to set afterburner to %d: %s\n", |
258 | s->afterburner, aac_get_error(err)); |
259 | goto error; |
260 | } |
261 | |
262 | if (avctx->cutoff > 0) { |
263 | if (avctx->cutoff < (avctx->sample_rate + 255) >> 8 || avctx->cutoff > 20000) { |
264 | av_log(avctx, AV_LOG_ERROR, "cutoff valid range is %d-20000\n", |
265 | (avctx->sample_rate + 255) >> 8); |
266 | goto error; |
267 | } |
268 | if ((err = aacEncoder_SetParam(s->handle, AACENC_BANDWIDTH, |
269 | avctx->cutoff)) != AACENC_OK) { |
270 | av_log(avctx, AV_LOG_ERROR, "Unable to set the encoder bandwidth to %d: %s\n", |
271 | avctx->cutoff, aac_get_error(err)); |
272 | goto error; |
273 | } |
274 | } |
275 | |
276 | if ((err = aacEncEncode(s->handle, NULL, NULL, NULL, NULL)) != AACENC_OK) { |
277 | av_log(avctx, AV_LOG_ERROR, "Unable to initialize the encoder: %s\n", |
278 | aac_get_error(err)); |
279 | return AVERROR(EINVAL); |
280 | } |
281 | |
282 | if ((err = aacEncInfo(s->handle, &info)) != AACENC_OK) { |
283 | av_log(avctx, AV_LOG_ERROR, "Unable to get encoder info: %s\n", |
284 | aac_get_error(err)); |
285 | goto error; |
286 | } |
287 | |
288 | avctx->frame_size = info.frameLength; |
289 | avctx->initial_padding = info.encoderDelay; |
290 | ff_af_queue_init(avctx, &s->afq); |
291 | |
292 | if (avctx->flags & AV_CODEC_FLAG_GLOBAL_HEADER) { |
293 | avctx->extradata_size = info.confSize; |
294 | avctx->extradata = av_mallocz(avctx->extradata_size + |
295 | AV_INPUT_BUFFER_PADDING_SIZE); |
296 | if (!avctx->extradata) { |
297 | ret = AVERROR(ENOMEM); |
298 | goto error; |
299 | } |
300 | |
301 | memcpy(avctx->extradata, info.confBuf, info.confSize); |
302 | } |
303 | return 0; |
304 | error: |
305 | aac_encode_close(avctx); |
306 | return ret; |
307 | } |
308 | |
309 | static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, |
310 | const AVFrame *frame, int *got_packet_ptr) |
311 | { |
312 | AACContext *s = avctx->priv_data; |
313 | AACENC_BufDesc in_buf = { 0 }, out_buf = { 0 }; |
314 | AACENC_InArgs in_args = { 0 }; |
315 | AACENC_OutArgs out_args = { 0 }; |
316 | int in_buffer_identifier = IN_AUDIO_DATA; |
317 | int in_buffer_size, in_buffer_element_size; |
318 | int out_buffer_identifier = OUT_BITSTREAM_DATA; |
319 | int out_buffer_size, out_buffer_element_size; |
320 | void *in_ptr, *out_ptr; |
321 | int ret; |
322 | AACENC_ERROR err; |
323 | |
324 | /* handle end-of-stream small frame and flushing */ |
325 | if (!frame) { |
326 | in_args.numInSamples = -1; |
327 | } else { |
328 | in_ptr = frame->data[0]; |
329 | in_buffer_size = 2 * avctx->channels * frame->nb_samples; |
330 | in_buffer_element_size = 2; |
331 | |
332 | in_args.numInSamples = avctx->channels * frame->nb_samples; |
333 | in_buf.numBufs = 1; |
334 | in_buf.bufs = &in_ptr; |
335 | in_buf.bufferIdentifiers = &in_buffer_identifier; |
336 | in_buf.bufSizes = &in_buffer_size; |
337 | in_buf.bufElSizes = &in_buffer_element_size; |
338 | |
339 | /* add current frame to the queue */ |
340 | if ((ret = ff_af_queue_add(&s->afq, frame)) < 0) |
341 | return ret; |
342 | } |
343 | |
344 | /* The maximum packet size is 6144 bits aka 768 bytes per channel. */ |
345 | if ((ret = ff_alloc_packet2(avctx, avpkt, FFMAX(8192, 768 * avctx->channels), 0)) < 0) |
346 | return ret; |
347 | |
348 | out_ptr = avpkt->data; |
349 | out_buffer_size = avpkt->size; |
350 | out_buffer_element_size = 1; |
351 | out_buf.numBufs = 1; |
352 | out_buf.bufs = &out_ptr; |
353 | out_buf.bufferIdentifiers = &out_buffer_identifier; |
354 | out_buf.bufSizes = &out_buffer_size; |
355 | out_buf.bufElSizes = &out_buffer_element_size; |
356 | |
357 | if ((err = aacEncEncode(s->handle, &in_buf, &out_buf, &in_args, |
358 | &out_args)) != AACENC_OK) { |
359 | if (!frame && err == AACENC_ENCODE_EOF) |
360 | return 0; |
361 | av_log(avctx, AV_LOG_ERROR, "Unable to encode frame: %s\n", |
362 | aac_get_error(err)); |
363 | return AVERROR(EINVAL); |
364 | } |
365 | |
366 | if (!out_args.numOutBytes) |
367 | return 0; |
368 | |
369 | /* Get the next frame pts & duration */ |
370 | ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts, |
371 | &avpkt->duration); |
372 | |
373 | avpkt->size = out_args.numOutBytes; |
374 | *got_packet_ptr = 1; |
375 | return 0; |
376 | } |
377 | |
378 | static const AVProfile profiles[] = { |
379 | { FF_PROFILE_AAC_LOW, "LC" }, |
380 | { FF_PROFILE_AAC_HE, "HE-AAC" }, |
381 | { FF_PROFILE_AAC_HE_V2, "HE-AACv2" }, |
382 | { FF_PROFILE_AAC_LD, "LD" }, |
383 | { FF_PROFILE_AAC_ELD, "ELD" }, |
384 | { FF_PROFILE_UNKNOWN }, |
385 | }; |
386 | |
387 | static const AVCodecDefault aac_encode_defaults[] = { |
388 | { "b", "0" }, |
389 | { NULL } |
390 | }; |
391 | |
392 | static const uint64_t aac_channel_layout[] = { |
393 | AV_CH_LAYOUT_MONO, |
394 | AV_CH_LAYOUT_STEREO, |
395 | AV_CH_LAYOUT_SURROUND, |
396 | AV_CH_LAYOUT_4POINT0, |
397 | AV_CH_LAYOUT_5POINT0_BACK, |
398 | AV_CH_LAYOUT_5POINT1_BACK, |
399 | #ifdef AACENCODER_LIB_VL0 |
400 | AV_CH_LAYOUT_7POINT1_WIDE_BACK, |
401 | AV_CH_LAYOUT_7POINT1, |
402 | #endif |
403 | 0, |
404 | }; |
405 | |
406 | static const int aac_sample_rates[] = { |
407 | 96000, 88200, 64000, 48000, 44100, 32000, |
408 | 24000, 22050, 16000, 12000, 11025, 8000, 0 |
409 | }; |
410 | |
411 | AVCodec ff_libfdk_aac_encoder = { |
412 | .name = "libfdk_aac", |
413 | .long_name = NULL_IF_CONFIG_SMALL("Fraunhofer FDK AAC"), |
414 | .type = AVMEDIA_TYPE_AUDIO, |
415 | .id = AV_CODEC_ID_AAC, |
416 | .priv_data_size = sizeof(AACContext), |
417 | .init = aac_encode_init, |
418 | .encode2 = aac_encode_frame, |
419 | .close = aac_encode_close, |
420 | .capabilities = AV_CODEC_CAP_SMALL_LAST_FRAME | AV_CODEC_CAP_DELAY, |
421 | .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S16, |
422 | AV_SAMPLE_FMT_NONE }, |
423 | .priv_class = &aac_enc_class, |
424 | .defaults = aac_encode_defaults, |
425 | .profiles = profiles, |
426 | .supported_samplerates = aac_sample_rates, |
427 | .channel_layouts = aac_channel_layout, |
428 | }; |
429 |